What a review is
- A public read, and a list of cases
- QA-lab URL Review fetches a public https address, describes the response, and proposes test cases. It does not run those cases, sign in, submit forms, or generate load.
- What the report scores
- Six areas from 0 to 100: SEO, GEO (whether AI search and answer engines can read and cite the site), analytics, security, accessibility, and performance. Each check is marked fails now, at risk, passes, or needs a person, with the evidence, a fix, and an effort estimate. You also see how the site looks in a Google result and a shared link, which AI crawlers its robots.txt lets in, a map of the pages read, and what to fix first. A headless browser loads up to three pages from an EU address: it records tracking before and after a cookie choice, runs axe-core, compares the HTML text with the rendered page, checks phone and tablet widths, and keeps three screenshots.
- Frameworks and tools
- Playwright, Cypress, Robot, Selenium, WebdriverIO, CodeceptJS, Selenide, TestNG, Watir, Cucumber, Behave, Reqnroll, Jest, Vitest, Mocha, Appium, Maestro, JMeter, Gatling, and k6 are the frameworks. axe-core runs in the review's headless browser on up to three pages, and Lighthouse runs on the start page through Google PageSpeed Insights. BrowserStack, Applitools, Percy, Provar, and Copado are named when they apply and are not launched against your site.
- Automation scripts
- It proposes test cases from one public https address and writes each one for the frameworks your team uses, from Playwright and Cypress to Selenide, Cucumber, Jest, Appium, and k6. The case export includes a starting script per framework and a ZIP with one suite each, a GitHub Actions workflow that runs them, and a fix kit with header files for common hosts, a Report-Only content security policy, and drafts for structured data, llms.txt, security.txt, and DNS records. The scripts are yours to run; the review never runs them.
- Your site
- It only reads public pages. It does not sign in, submit forms, or generate load, and it does not run the cases it proposes.
- How far it reads
- Your first review is included: up to twenty pages, two links deep. Want a wider or deeper look? You see the price before anything starts.
- Where cases go
- Automation scripts as a ZIP, Excel, TestRail, Xray, Zephyr Scale, Jira, Azure Test Plans, MantisBT, JSON, plain text, or PDF. Every format carries the scripts for your frameworks. Case export is priced on the page.
- Language
- The page is written in English and can be read in Spanish, Portuguese, German, French, and Italian. The news strip follows the country of the connection and the language of the page.
- Rules
- A review reads a public page, tells you what it found, and suggests test cases your team can use. The disclaimers explain how that works, including what applies in the country of the site you are reviewing.
- Elsewhere
- Films and notes on YouTube, TikTok, X, LinkedIn, DEV, and Reddit.
The request
Who is asking, and which address should we open?
Review log
Lines appear here while the pages are being read. Stop keeps the pages already saved and writes that stop into the report.